Files
nick-tracker/GUIDE.md
2026-01-11 07:51:30 +00:00

4.4 KiB

nick-tracker - Implementation Guide

Overview

This guide walks you through implementing nick-tracker using the Ralph Method with phase-by-phase execution.

Tech Stack: TypeScript / NestJS 10 Features: 10 Estimated Cost: ~$5.00 (at ~$0.50/feature)

Prerequisites

  1. Read PROMPT.md for full project requirements
  2. Review prd.json for feature tracking
  3. Ensure your environment is set up for TypeScript

Execution

Execute each phase in order. Do not proceed to the next phase until the current phase is complete.

Phase 1: Foundation

/ralph-wiggum:ralph-loop "$(cat prompts/phase1-prompt.txt)" --max-iterations 30 --completion-promise "PHASE_1_COMPLETE"

Verify:

npm run build && npm run test && npm run lint

Bash loop alternative (if plugin unavailable):

PROMPT=$(cat prompts/phase1-prompt.txt)
MAX_ITERATIONS=30
COMPLETION_PROMISE="PHASE_1_COMPLETE"

for i in $(seq 1 $MAX_ITERATIONS); do
  echo "=== Iteration $i of $MAX_ITERATIONS ==="
  RESPONSE=$(claude -p "$PROMPT")
  echo "$RESPONSE"
  if echo "$RESPONSE" | grep -q "$COMPLETION_PROMISE"; then
    echo "Phase 1 complete!"
    break
  fi
  if echo "$RESPONSE" | grep -q "ABORT_BLOCKED"; then
    echo "Blocked - manual intervention required"
    break
  fi
done

Phase 2: Core

/ralph-wiggum:ralph-loop "$(cat prompts/phase2-prompt.txt)" --max-iterations 40 --completion-promise "PHASE_2_COMPLETE"

Verify:

npm run build && npm run test && npm run lint

Bash loop alternative:

PROMPT=$(cat prompts/phase2-prompt.txt)
MAX_ITERATIONS=40
COMPLETION_PROMISE="PHASE_2_COMPLETE"

for i in $(seq 1 $MAX_ITERATIONS); do
  echo "=== Iteration $i of $MAX_ITERATIONS ==="
  RESPONSE=$(claude -p "$PROMPT")
  echo "$RESPONSE"
  if echo "$RESPONSE" | grep -q "$COMPLETION_PROMISE"; then
    echo "Phase 2 complete!"
    break
  fi
  if echo "$RESPONSE" | grep -q "ABORT_BLOCKED"; then
    echo "Blocked - manual intervention required"
    break
  fi
done

Phase 3: Integration

/ralph-wiggum:ralph-loop "$(cat prompts/phase3-prompt.txt)" --max-iterations 40 --completion-promise "PHASE_3_COMPLETE"

Verify:

npm run build && npm run test && npm run lint

Bash loop alternative:

PROMPT=$(cat prompts/phase3-prompt.txt)
MAX_ITERATIONS=40
COMPLETION_PROMISE="PHASE_3_COMPLETE"

for i in $(seq 1 $MAX_ITERATIONS); do
  echo "=== Iteration $i of $MAX_ITERATIONS ==="
  RESPONSE=$(claude -p "$PROMPT")
  echo "$RESPONSE"
  if echo "$RESPONSE" | grep -q "$COMPLETION_PROMISE"; then
    echo "Phase 3 complete!"
    break
  fi
  if echo "$RESPONSE" | grep -q "ABORT_BLOCKED"; then
    echo "Blocked - manual intervention required"
    break
  fi
done

Phase 4: Polish

/ralph-wiggum:ralph-loop "$(cat prompts/phase4-prompt.txt)" --max-iterations 30 --completion-promise "PHASE_4_COMPLETE"

Verify:

npm run build && npm run test && npm run lint

Bash loop alternative:

PROMPT=$(cat prompts/phase4-prompt.txt)
MAX_ITERATIONS=30
COMPLETION_PROMISE="PHASE_4_COMPLETE"

for i in $(seq 1 $MAX_ITERATIONS); do
  echo "=== Iteration $i of $MAX_ITERATIONS ==="
  RESPONSE=$(claude -p "$PROMPT")
  echo "$RESPONSE"
  if echo "$RESPONSE" | grep -q "$COMPLETION_PROMISE"; then
    echo "Phase 4 complete!"
    break
  fi
  if echo "$RESPONSE" | grep -q "ABORT_BLOCKED"; then
    echo "Blocked - manual intervention required"
    break
  fi
done

Full Project Execution (Alternative)

If you prefer to run the entire project in one loop:

/ralph-wiggum:ralph-loop "$(cat PROMPT.md)" --max-iterations 100 --completion-promise "PROJECT_COMPLETE"

Note: Phase-by-phase execution is recommended for complex projects as it provides better control and verification checkpoints.

Tracking Progress

  • Check prd.json to see which features have passes: true
  • Review progress.txt for the implementation log
  • All phases complete when all features pass

Troubleshooting

If the agent outputs ABORT_BLOCKED:

  1. Review the error message
  2. Fix the blocking issue manually
  3. Re-run the current phase

If iterations exhaust without completion:

  1. Check progress.txt for what was accomplished
  2. Review prd.json for remaining features
  3. Increase --max-iterations and re-run

Generated with Ralph PRD Generator